RoSETTA: Inverting coastal landforms towards past sea-levels, tectonics, and coral reef dynamics

Coastal landforms constitute a worldwide archive that intricately records past sea-level, morphotectonics and (bio-)morphogenesis. Although forward landscape evolution modelling has shown its potential in deciphering, independently, either sea-level, tectonics, or morphogenesis, this strategy rests upon a wealth of hypotheses regarding the other 2 aspects. We circumvent this limitation by inverting coral reef terraces geometries using a Monte-Carlo Markov Chain sampling in a Bayesian framework with the code RoSETTA (Resolving Sea-level, Ecosystems and Tectonics from Terrace Analysis). Probabilistic solutions of past sea-level, as well as vertical motion, erosion and reef growth rates are simultaneously obtained from reconstructions of stacks of coastal landforms. We first benchmark RoSETTA on synthetic sequences of marine terraces and then on the canonical landforms of Sumba island, Indonesia. Beyond successful reconstructions, our probablistic approach allows to highlight periods of time and sea level that can confidently be resolved from a given stack of marine terraces. Additionally, we can also point out periods for which the archive is insufficient to be conclusive, an overlooked aspect of earlier attempts to decipher coastal landscapes. A successful remedy is to use joint inversions of multiple profiles using RoSETTA, improving the confidence of the results.